{"id":"arxiv/2610.38496","url":"https://arxiv.org/abs/2610.38496","tags":["cs.AI","cs.LG","math.OC"],"score":0,"title":"Accelerating Nonlinear MPC by Behavior Cloning with Separated Timescales: Application to Nuclear Reactor Load-Following","domain":"cs.AI","source":"arxiv","authors":["Perceval Beja-Battais","Alain Grosset{ê}te","Nicolas Vayatis"],"reviews":0,"subject":"cs.AI","summary":"Behavior Cloning (BC) of expert controllers for multiscale dynamical systems is challenging: state variables evolving at widely separated timescales induce highly nonlinear couplings that a shared latent space may fail to adequately represent, degrading closed-loop generalization. We propose Structured Representation Learning for Behavior Cloning (SRLBC), an architecture designed to leverage the physical knowledge of multiscale systems in which slow state variables, fast (stiff) state variables, and predictable exogenous inputs evolve at different timescales. On 500 long-horizon roll-outs of…","category":"research","comments":0,"disputed":false,"division":"cs","mentions":0,"discussed":0,"published":"2026-10-28","reactions":{}}